Is this photo real?
Upload any photo. We tell you if it’s Original, Photoshopped or AI. If it’s AI, it gets popped!
Drop a photo here
JPG, PNG, WEBP up to 10MB · or paste with Ctrl/⌘+V
Quick Test Samples
no photo handy? try one of ours ↓
- Answer in about a second
- Checked in memory — your photo is never stored
- WhatsApp-forwarded photos are never counted against you
Built in India for the refund desk, the HR inbox and anyone who’s ever thought “hmm, is this real?”
- 50–70%
- of disputed refunds Zomato absorbs
- No. 1
- fastest-growing return scam (Forter)
- 1 lakh+
- fake degrees seized in Kerala
How it works
Three simple steps
…pretty much how popcorn is made.
- 1
Drop in your kernel
Your photo goes in like a corn kernel — a refund photo, listing, certificate or claim. WhatsApp-forwarded photos are fine.
— no sign-up, nothing saved
- 2
We turn up the heat
Hidden metadata, edited spots, copied regions, watermarks and three AI detectors all get to work at once — in about a second.
— 12 checks, one go
- 3
Fakes go pop!
A real photo stays a calm kernel: Original. Edits get flagged as Photoshopped. AI images pop — with the reasons and a heatmap.
— you see exactly why
Layer 1
Metadata & provenance
Signatures of 60+ editing apps and AI generators (ChatGPT, Gemini, Midjourney, PicsArt…), C2PA Content Credentials, camera-thumbnail checks and a photo fingerprint to catch repeat submissions.
Layer 2
Edited-spot forensics
Error Level Analysis, clone-stamp detection and sensor-noise checks reveal pasted text, copied regions, swapped seals and healed-out areas.
Layer 3
AI detector ensemble
Modern deep-learning detectors trained on GPT-Image, Nano Banana, Midjourney, FLUX and Seedream images, robust to WhatsApp compression.
Built for every photo that costs you money
- Amazon & Flipkart returns
- Zepto & Instamart refunds
- Zomato & Swiggy complaints
- IndiaMART listings
- Property photos
- Certificates for HR
- School admissions
- Insurance claims
Honest customers don’t get flagged: WhatsApp compression and stripped metadata are never counted against a photo.
Pricing
Cheaper than one fake refund
Pay as you go
- Simple REST API
- Verdict, heatmap & reasons
- Photo fingerprint for repeat scammers
Monthly plans for big apps
- Volume pricing from 1 lakh checks
- Uptime SLA & priority support
- On-prem / VPC deployment
Free for everyone on our website
- Check any photo here
- No sign-up needed
- Photos never stored
One fake refund (₹1,820) pays for 1,820 checks.
Checking 50,000 photos costs ₹50,000/month. It pays for itself after catching 28 fake refunds — just 0.06% of photos.
For business
Plug it in with 5 lines of code
Built for engineering teams at quick-commerce, food delivery, marketplaces and HR platforms. Send a photo, get back a verdict your support team can act on — no ML team needed. Photos are processed in memory and never stored (DPDP-ready).
FAQ
AI photo detection, answered
How do I check if a photo is AI-generated?
Upload the photo to Popped! (or paste it with Ctrl/⌘+V). In about a second you get one of three answers — Original, Photoshopped or AI = Popped! — with plain-English reasons, a tamper heatmap and a list of every check we ran. It's free on the website.
Can Popped! detect ChatGPT, Gemini (Nano Banana) and Midjourney images?
Yes. Three deep-learning detectors trained on 2025–26 generators — including GPT-Image, Gemini / Nano Banana Pro, Midjourney 7, Seedream, FLUX and Grok — work together, and Popped! also reads AI tags, Content Credentials (C2PA) and Gemini's visible sparkle watermark. In our evaluation the ensemble caught 87% of AI images at a 1% false-alarm rate.
How does Popped! know a photo was Photoshopped?
It looks for editing-app signatures (Photoshop, PicsArt, Canva, Snapseed, PixelLab and 40+ more), Error Level Analysis anomalies where text or seals were pasted, clone-stamped regions that duplicate camera noise, and camera thumbnails that no longer match the photo.
Will a WhatsApp-forwarded photo be marked fake?
No. WhatsApp strips metadata and re-compresses every photo, so Popped! treats missing EXIF and WhatsApp compression as neutral. On 142 real smartphone photos, also re-sent through WhatsApp-style compression, only 1 was flagged.
Is my photo stored or used for training?
No. Photos are analysed only in memory and discarded as soon as the answer is sent. They are never written to disk, logged, shared or used to train models.
Is the AI detector 100% accurate?
No detector is. Popped! is a fast screening tool: treat an AI = Popped! or Photoshopped result as a reason for a closer human look, not as proof on its own.
Is there an API for businesses?
Yes. Food-delivery, e-commerce, quick-commerce, insurance and HR platforms can call POST /api/v1/check from their refund or onboarding flow at ₹1 per photo, with monthly plans for high volume.